Official implementation of GPS-GLASS: Learning Nighttime Semantic Segmentation Using Daytime Video and GPS data arXiv
| ACDC-night | Dark Zurich-val |
|---|---|
- python3.7
- pytorch==1.9.0
- cuda10.2
Cityscapes: Please follow the instructions in Cityscape to download the training set.
Dark-Zurich: Please follow the instructions in Dark-Zurich to download the training/val/test set.
ACDC: Please follow the instructions in ACDC to download the training/val/test set.
NightCity+: Please follow the instructions in NightCity+ to download the training/val set.
Pretrained models can be downloaded form here.
To reproduce the reported results in our paper, follow these steps:
Step1: download the trained models and put it in the root.
Step2: change the data and model paths in configs/test_config.py
Step3: run "python evaluate.py"
Step4: change the ground truth data path in compute_iou.py
Step5: run "python compute_iou.py"
The code is based on DANNet, AdaptSegNet .
@InProceedings{lee2023gps,
author = {Lee, Hongjae and Han, Changwoo and Jung, Seung-Won},
title = {{GPS-GLASS}: Learning Nighttime Semantic Segmentation Using Daytime Video and GPS data},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops},
year = {2023}
}